jaroslav hajek and asymptotic theory of rank tests - Kybernetika
jaroslav hajek and asymptotic theory of rank tests - Kybernetika
jaroslav hajek and asymptotic theory of rank tests - Kybernetika
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
246 J. JURECKOVA<br />
6. FURTHER ASYMPTOTIC PROPERTIES OF RANKS<br />
Hajek [9] demonstrated that not only the best Pitman efficiency but also the best<br />
exact Bahadur slope is attainable by <strong>rank</strong> statistics; otherwise speaking, that the<br />
vector <strong>of</strong> <strong>rank</strong>s is sufficient in the Bahadur sense.<br />
Consider the two-sample model with independent samples X\,..., X n <strong>and</strong> Y\,...<br />
. .. ,Y m with the respective densities <strong>and</strong> d.f.'s /, g, F, G; let limAr^oo w+n = \ £<br />
(0,1) <strong>and</strong> denote<br />
H(x) = \F(x) + (l-\)G(x), xeRi, (42)<br />
J(u)=-^-F(H- l (u)), g( u )=±G(H- 1 (u)), 0